... | ... | @@ -57,12 +57,16 @@ Performance goals: |
|
|
|
|
|
- SumSquares \[FINE\]
|
|
|
- Dot product \[FINE\]
|
|
|
- Evens **\[OK, but more than 3 times slower than C\]****\[ROMAN (might affect Quickhull due to packByTags\]**
|
|
|
- Evens '\[OK (but more than 3 times slower than C)\]'
|
|
|
|
|
|
- rl reckons this is due to GHC compiling modulo of powers of two inefficiently; c.f., [\#3065](https://gitlab.haskell.org//ghc/ghc/issues/3065) (in `packByTags`)
|
|
|
- SMVM **\[SLOW (lack of fusion)\]****\[BEN & ROMAN\]**
|
|
|
- Dynamically-nested DPH programs without user-defined datatypes should run correctly, but not necessarily fast
|
|
|
|
|
|
- Quicksort **\[BROKEN ([SpecConstr](spec-constr)) & SLOW\]****\[SIMON & BEN\]**
|
|
|
- Quickhull **\[OK, but has a [SpecConstr](spec-constr) problem that we want to fix\]****\[ROMAN\]**
|
|
|
|
|
|
- Probably affected by the same optimisation issue with the compilation of modulo operations as Evens
|
|
|
- Dynamically-nested DPH programs with user-defined datatypes should run correctly, but not necessarily fast
|
|
|
|
|
|
- Words **\[BROKEN ([SpecConstr](spec-constr))\]****\[SIMON\]**
|
... | ... | |